ADP-ribosylation factors (ARFs) are approximately 20 kDa guanine nucleotide-binding proteins initially identified by their ability to enhance in vitro cholera toxin-catalyzed ADP-ribosylation and subsequently shown to participate in vesicular transport in the Golgi and other cellular compartments. ARFs are active in the GTP-bound form; hydrolysis of bound GTP to GDP, possibly with the assistance of a GTP hydrolysis (GTPase)-activating protein (GAP) results in inactivation. Exchange of GDP for GTP and reactivation were shown by other workers to be enhanced by Golgi membranes in a brefeldin A (BFA)-sensitive reaction, leading to the proposal that the guanine nucleotide-exchange protein was a BFA target. In these studies, a guanine nucleotide-exchange protein (GEP) for ARF has been highly purified from rat spleen cytosol in a brefeldin A- insensitive form. GEP appeared to be an approximateky 55-kDa protein that represents only approximately 0.0006% of the cytosolic protein. The activity was maximal with 1.5 to 5 mM MgCl2 (in the presence of 1 mM EDTA). At slightly lower concentrations of MgCl2, uncatalyzed nucleotide binding was faster and no GEP effect was seen. With GEP, >80% of ARF1 or ARF3 bound GTPgammaS in 40 minutes. The rate of binding of [3H]GDP was much slower. PI, PA, or PS (200 muM) was necessary for GEP activity; half maximal activation was seen with 15-20 muM. PE, PC, DMPC/cholate and PIP2 were ineffective. GEP stimulated the release of bound [35S]GTPgammaS from ARF3 only in the presence of unlabeled GTPgammaS, GTP, or GDP. Replacement by GTPgammaS was faster and occurred at a lower nucleotide concentration than that by GDP. As GEP substrates, Class I ARFs were much better than those of Class II or Class III and myristoylation of the N-terminal glycine was clearly important.